Transformation and Synthesis of Artifact-centric Business Processes

In today’s dynamic business environment, business process modeling (BPM) has become a fundamental tool in many organizations to gain operational benefits and stay competitive with their rivals. Thus, there is always an increasing demand for modelling paradigms to support flexible and reusable process design, facilitate the accommodation of rapidly changing business requirements, and provide insights on process progress to improve process performance and minimize operating costs. The activity-centric and artifact-centric approaches are the two major modelling paradigms in BPM. While the traditional activity-centric paradigm emphasises representing activities and their control flows, the recent artifact-centric paradigm provides equal support to control flow and data by incorporating “artifacts” which are key business entities and their lifecycles. In recent years, the artifact-centric paradigm has been extensively studied and there is evidence that this paradigm provides better flexibility and reusability support compared with the traditional activity-centric paradigm. With the prevalence of the artifact-centric paradigm, the transformation of traditional activity-centric process models into artifact-centric process models has emerged as an important research challenge. Most of the existing approaches for this purpose have a restricted view on the artifacts and their interaction (or synchronization) dependencies, and only support a semi-automatic transformation. Furthermore, these approaches focus on transforming standalone activity-centric process models that specify activities performed in one single organization, while ignoring the inter-organisational business process (IOBP) models that represent activities and data (artifacts) distributed among various organizations. In addition, the transformation of artifact-centric process models into activity-centric process models has also received considerable attention, as the activity-centric process models provide better process view compared with the artifactcentric process models that contain unorganized sets of business rules. The existing approaches to achieve this objective are also limited in terms of using artefact lifecycles rather than the artifact-centric process models to construct the activity-centric process models and only supporting a semi-automatic transformation. Therefore, automated approaches are required to improve the scope and efficiency of the proposed transformations. In this regard, this thesis aims to address the following research questions: (1) How to efficiently transform the activity-centric process models into the artifact-centric process models?; (2) How to merge the collaborating processes of activity-centric IOBP (inter-organizational business process) models?; and (3) How to construct the activity-centric process models from the artifact-centric process models. First, an efficient tree-based approach is proposed to transform activity-centric process models into artifact-centric process models. The proposed transformation mainly aims to synthesize or generate synchronized lifecycles of artifacts from the activity-centric process models. The proposed approach provides a set of algorithms that initially extracts the hierarchical tree representation of the activity-centric process model, then generates a lifecycle for every artifact, and also synchronizes the generated lifecycles. The proposed approach is demonstrated using a case study and also implemented and evaluated using a process model collection from the BPM Academic Initiative (BPMAI). Second, a process interaction-based approach is proposed to merge the collaborating processes of activity-centric inter-organizational business process (IOBP) models that have artifact annotations in order to synthesize the lifecycles of artifacts from the resulting integrated process models. The proposed approach comprises a set of algorithms that merge two or more collaborating processes based on their interaction patterns. Specifically, the approach first identifies the type of interaction, such as synchronous or asynchronous, between the collaborating processes and then merge them according to the rules formulated for each type of process interaction. The proposed approach is demonstrated using a case study and also implemented and evaluated using a model collection from the BPMAI. Last, a reverse transformation approach is proposed for constructing activity-centric process models from artifact-centric process models, and for checking the consistency between the constructed model and the base model. A trace-based method is used to analyze the model consistency, where the execution traces of the base model is analyzed over the execution traces of the constructed model. The proposed approach is demonstrated using a case study and also implemented and evaluated using two motivating process models